JPH1173726A - Signal processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To perform musical processing suited to a musical tune by selecting an optimum characteristic from plural beforehand prepared characteristics based on the musical tune of an inputted audio signal and compression-encoding it. SOLUTION: An input signal S1 is converted to a digital signal S2, and is band divided by plural filters of a musical tune analytic part 12, and the peak levels of the signals are detected respectively by a level detection circuit for every prescribed time width to be inputted to a CPU 14. The CPU 14 decides characteristics related to respective parameters used for encoding in an audio data high efficiency encoding system ATRAC encoder 10 based on the contents of respective detection signals to supply a selection signal Sc to the ATRAC encoder 10. The ATRAC encoder 10 selects the optimum characteristic used for encoding from plural beforehand prepared characteristics related to respective parameters based on the selection signal Sc to compression-encode the inputted digital audio signal S2. Thus, the characteristics used at an encoding time are revised every moment at need based on the musical tune of the inputted signal.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention
Disc, hereinafter referred to as “MD”. ), And more particularly, to signal processing when recording an audio signal on an MD.

[0002]

2. Description of the Related Art In an MD, an ATRAC (Adaptive) is used in order to compress information to about 1/5 of that of a compact disc (Compact Disc, hereinafter referred to as "CD").
It uses a high-efficiency audio data encoding method called TRansform Acoustic Coding. The idea in ATRAC is to take into account the human auditory characteristics and remove the redundant portion of the signal that cannot be perceived by humans and perform compression. In other words, on the assumption that a difference occurs between the original signal and the compressed signal (that is, quantization noise is generated), the compression is performed while controlling the noise so that a human cannot perceive it.

[0003] Specifically, in ATRAC, an equal loudness characteristic (minimum audible curve) related to human auditory properties
High-efficiency coding is performed using the masking effect and the masking effect.
This will be described with an example. In FIG. 7A, a curve 100 indicates a minimum audibility curve, and sounds at levels below this curve are not heard by human ears.
Now, assuming that the input signal includes the sound A and the sound B in the figure, the sound B is audible to the human ear, but the sound A is located below the minimum audible curve, so that the human ear does not. Inaudible.
Therefore, in ATRAC, this sound B is recorded as a valid signal, but sound A is not recorded (thinned out).
Alternatively, the encoding is performed with the quantization bit allocation extremely reduced.

[0004] FIG. 7B shows a masking characteristic 110 of the sound C in addition to the minimum audible curve 100. The sound B is located above the minimum audible characteristic 100, but is masked by the sound C in the vicinity thereof and cannot be heard by human ears. Therefore, ATRAC performs compression by removing sound B as a redundant part.

[0005]

As described above, the ATR
The method of signal compression in AC depends on how to set a minimum audible curve, a masking characteristic, and the like, which are parameters in the compression processing. In this regard, in the past, ATRAC encoding has generally been performed for each of these using only one predetermined characteristic.

However, depending on the setting of the minimum audible curve and the masking characteristics used in performing the compression coding by ATRAC, the recorded signal, that is, the feeling of music is considerably different. Therefore, when music of various genres and tunes is recorded using only one characteristic, the original sound (original music) is not always subjected to optimal compression encoding, and the recorded and reproduced music is impressive. It may not be possible to record in accordance with the music, for example, the user cannot feel the sound, or the high frequency band is harsh.

SUMMARY OF THE INVENTION The present invention has been made in view of the above points, and provides a music signal processing apparatus capable of performing compression encoding using optimal characteristics according to the tune of a music piece to be recorded. The task is to provide.

[0008]

According to a first aspect of the present invention, there is provided a signal processing apparatus for generating a spectrum signal from an input digital audio signal and compressing the generated spectrum signal. Encoding means for compression-encoding based on the characteristics of the parameters for the input digital audio signal, from the input digital audio signal, to determine the tune of the song composed of the signal, based on the determination result, a plurality of prepared in advance And a characteristic control unit for selecting one of the characteristics, wherein the encoding unit is configured to perform encoding using the characteristic selected by the characteristic control unit.

[0009] According to the signal processing device configured as described above, the input digital audio signal is input to the encoding means, converted into a spectrum signal, and then compression-coded based on characteristics for compression. On the other hand, the input digital audio signal is also input to the characteristic control means, and the tone of the music composed by the signal is determined. The characteristic control means selects an optimum characteristic for the tune from a plurality of characteristics prepared in advance for a certain parameter according to the determined tune. The encoding unit performs compression encoding suitable for the tune based on the characteristic selected by the tune control unit.

According to a second aspect of the present invention, in the signal processing device of the first aspect, the characteristic control means continuously determines a tune and selects the characteristic in real time.

[0011] According to the signal processing device configured as described above, the tune is determined in real time during the encoding of the input signal, and the characteristic is selected. That is, since the compression encoding is performed while the characteristics are appropriately changed according to the tune of the input audio signal, the encoding is achieved using the optimal characteristics by following the tune change even during one song. You.

According to a third aspect of the present invention, in the signal processing device according to the first or second aspect, the characteristic control means includes a band dividing means for dividing the input digital audio signal into a plurality of bands, and a band dividing means. And means for detecting the peak level of the signal of each band, and determining means for determining the tune based on the detected peak level.

According to the information processing apparatus configured as described above, the input digital audio signal is divided into a plurality of bands by the band dividing means, the peak level is detected for each band, and based on the detected peak level. The tune is determined.

According to a fourth aspect of the present invention, in the signal processing device according to any one of the first to third aspects, the parameter includes at least one of a minimum audible characteristic and a masking characteristic.

According to the signal processing device configured as described above, encoding by the encoder is executed based on the optimum minimum audible characteristic and masking characteristic selected according to the tune of the input audio signal.

[0016]

Preferred embodiments of the present invention will be described below with reference to the accompanying drawings.

[0017] The principle explanation First, will be briefly described the principles of the present invention. As described above, in ATRAC encoding, the parameters used in encoding, specifically, the minimum audible curve,
It is important to actually set the masking characteristics, the time width (sound group to be described later) when temporally separating the signal to be coded, and the like, and the feeling of the recorded music is considerably different depending on the setting. In view of this point, in the present invention, the tune of a music signal to be recorded is analyzed and determined in real time, and according to the result, a plurality of preset characteristics optimal for encoding the music signal are prepared for each parameter. It is configured to be used by selecting from the characteristics of. As a result, optimal compression encoding can be performed using optimal characteristics according to the type of music to be recorded and the tune.

MD Recording Apparatus Next, a preferred embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a signal processing system of the MD recording device 1 according to the embodiment of the present invention.

In the MD recording device 1, the MD disc 1
An input signal S1 to be recorded at 8 is input to an A / D converter 8, sampled and quantized at a sampling frequency of 44.1 kHz, 16 to 20 bits, and output as a digital signal S2. The digital signal S2 is sent to the ATRAC encoder 10 and the tune analysis unit 12.

FIG. 2 shows the configuration of the tune analysis unit 12. The tune analysis unit 12 has a low-pass filter 20 and a plurality of band-pass filters 21 and 22, and divides the digital signal S2 input from the A / D converter 8 into bands. A preferable example of the band to be divided is three bands that are band-divided in the ATRAC encoder 10 (that is, low band: 0 to about 5.5 kHz, middle band: about 5.5 to 11 kHz, and high band: about 11 to 22 kHz). It is to match. This is because, within the ATRAC encoder, the signal is band-divided into these three bands, and compression encoding is performed by applying a minimum audible curve or the like. However, it is also possible to divide the frequency band into three or more finer bands and analyze each finer frequency band. Also,
As each filter, QMF (Quadrature Mirror Filt
er) is preferably used. This is because an aliasing component generated by band division can be canceled at the time of band synthesis. However, other filter configurations can be used.

The signals thus band-divided are input to level detection circuits 23, 24 and 25, respectively. The level detection circuits 23, 24, and 25 detect peak (maximum) levels in the input signal for each predetermined time width, and
Send to PU14. At this time, the time width for detecting the peak level is preferably one sound group (also referred to as a coded block, a maximum of 11.6 msec) which is a minimum unit for time-dividing the music signal in the ATRAC encoder 10. The reason is that the ATRAC encoder 10 at the subsequent stage compresses and encodes the signal in this unit, so that it is appropriate to analyze the tune in the same unit. However, if it is not possible to perform analysis in units of one sound group due to the time required for communication between the microcomputers and LSIs, it is necessary to determine the tune in units of the time width corresponding to several sound groups. Do.

The CPU 14 includes the level detection circuits 23, 2
4 and 25. Based on the content of the detection signals, characteristics of each parameter (bit allocation and time width of sound group, minimum audible curve, masking characteristic) to be used in encoding in the ATRAC encoder 10 are obtained. The selection signal Sc for determining and selecting an appropriate characteristic is set to ATR.
Supply to AC encoder 10.

The CPU 14 determines from the transition of the peak level signal for each band transmitted from each level detection circuit in a certain time range, for example, that there are many low-level high-frequency signals and that the level change of the middle-frequency signal is small. The tune is determined, and the bit allocation and time width, the minimum audibility characteristic and the masking characteristic of the sound group considered to be optimal in each case are determined. The details of the determination method will be described later.

The ATRAC encoder 10 is prepared in advance for each parameter with a plurality of characteristics to be used at the time of encoding, and is configured so that they can be selected by an external control signal Sc. The following are prepared in advance as a plurality of characteristics for each parameter. First, regarding the time width of the sound group, A
In addition to the maximum value of 11.6 msec specified on the TRAC, a plurality of shorter time widths can be selected. Also, the bit allocation of the sound group can be arbitrarily determined. As the minimum audibility curve, a plurality of characteristics such as a curve having a high level generally, a curve having a low level generally, and a curve having a slightly higher (or lower) high-frequency side are prepared. As the masking characteristic, a steep characteristic (that is, a nearby signal is hard to be masked) and a gentle characteristic (a nearby signal is easily masked) are prepared. Then, the control signal Sc supplied from the CPU 14
, The most appropriate one for each parameter is selected and compression-encoded. Therefore, based on the tune of the input signal, the ATRAC encoder 10
The characteristics used within will change from time to time as needed.

The signal S3 encoded in this way is stored in the recording unit 1
6 and recorded on the MD disk 18.

Next, a description will be given of a method of determining characteristics of each parameter performed by the music analysis unit 12 and the CPU 14. The characteristics that can be changed in the ATRAC encoder 10 are the bit allocation and time width of the sound group, the minimum audibility curve, and the masking characteristics, as described above. These will be described in order.

(1) Bit Allocation and Time Width of Sound Group In ATRAC, an input signal to be coded is firstly:
It is divided into a predetermined time width. One unit of the divided time window is called a sound group (or coded block), and is defined as a maximum of 11.6 msec on ATRAC. Therefore, encoding can be performed using a plurality of time widths in a range equal to or less than the maximum value. Also, the distribution of the number of quantization bits for each sound group can be arbitrarily determined.

FIG. 3 shows an example in which the bit allocation of the sound group is changed. FIG. 3A shows the case where the bit allocation is small, and FIG. 3B shows the case where the bit allocation is large. Increasing the bit allocation enables encoding with less quantization noise.

FIG. 4 shows an example in which the time width of the sound group is changed, and FIG. 4B shows a time width smaller than that of FIG. 4A. For this reason, the bit allocation amount in the level direction increases, and encoded (recorded) data at a level closer to the original sound is obtained. Therefore, for example, in the case of classical music, when the lingering part of a song is regarded as important, by setting this time width short, it is possible to make the lingering part a sound that is faithful to the original sound and has little distortion. .

(2) Minimum audible curve The minimum audible curve is a curve in which the sound pressure level audible to human ears is plotted against frequency. That is, it is assumed that a sound whose level is lower than this curve is inaudible to human ears. The ATRAC encoder prepares a plurality of curves and selects one of them to perform compression encoding. At the time of encoding, basically, a sound (signal) having a lower level than this curve is removed or encoded with a considerably smaller number of quantization bits. Note that, according to the minimum audibility curve or the masking characteristic described later, the term “not coded” includes both signals that are not coded and signals that are not coded with an appropriate number of quantization bits.

FIGS. 5A and 5B show a case where the same signal (ie, the same spectrum) is encoded using two different minimum audibility curves. The curve in FIG. 3A is a minimum audibility curve with a high level, and the curve in FIG. 5B is a minimum audibility curve with a low level. As can be seen, a signal that is not encoded using the curve of FIG. 5A will be encoded using the curve of FIG. 5B. In ATRAC, compression encoding is performed after an input signal is divided into three bands of a low band, a middle band, and a high band. Therefore, if a minimum audible curve used for a signal in any band is set low, the Many signals in the band can be recorded. For example, in normal music, a high-frequency signal generally has a lower signal level (contains many low-level signals) than a middle-frequency signal and a low-frequency signal. Therefore, if you want to emphasize high-frequency signals to some extent, such as rock music, select a low audible curve with a low level as the minimum audible curve used to encode the high-frequency signals, so that rich recording of high frequencies can be achieved. It can be carried out. If the high-frequency signal is not particularly important for classical music, etc., set the minimum audible curve of the high frequency higher, and instead set the lower curve of the middle frequency lower, and set the middle frequency curve lower. The thickness can be increased.

The tune of the rock type or the classic tune is determined by the data received from the tune analysis unit 12 by the CPU 14. For example, when a signal having a relatively high peak level appears periodically in a high frequency range, it is determined that the tone is a rock-based tune, and a signal having a high level is continuously averaged in the middle frequency range without a sharp change. If it appears, it can be determined that it is a classic tune. These are merely examples, and the tune can be determined based on various criteria.

(3) Masking Characteristic The masking characteristic means that when a sound C has a sound D of a higher level at a frequency close to that of the sound C, the sound C cannot be heard due to the presence of the sound D. (See FIG. 6A). FIGS. 6A and 6B show the case where different masking characteristics are applied to the same spectrum. The masking characteristic in FIG. 6A is a relatively gentle characteristic, and the characteristic in FIG. 6B is a relatively steep characteristic. As shown in the figure, if the gradual characteristic shown in FIG. 6A is used, the sound C is not encoded.
When the steep characteristic of (B) is used, the sound C is also encoded. Therefore, in the case where a relatively high-level signal frequently appears as in rock music, for example, FIG.
In some cases, it is better to make the masking characteristics steep to make it difficult to mask sounds near high-level signals.

As described above, in the present invention,
The tone of the signal to be recorded on the MD is determined, and based on the result, an optimum one of a plurality of parameters (minimum audibility, masking, etc.) used in ATRAC is selected and changed. It is composed.
The determination of the tune and the selection and change of the characteristic are preferably performed continuously in units of a predetermined time, which is preferably one sound group. Therefore, even if the tune changes during a single song, the characteristics will change every moment so as to follow the change, so that encoding and recording suitable for the tune of each song can be performed. Becomes

[0035]

According to the first aspect of the present invention, the characteristic control means selects an optimum one of a plurality of characteristics based on the tune of the input audio signal, and the encoding means selects the optimum characteristic based on the selected characteristics. Performs compression encoding of the input audio signal.
Therefore, compression encoding can be performed using characteristics suitable for the tune of the input signal.

According to the second aspect of the present invention, since the compression encoding is performed while the characteristics are appropriately changed in accordance with the tune of the input audio signal, the change in the tune can be optimally tracked even in a single piece of music. Encoding is achieved using various characteristics.

According to the third aspect of the present invention, since the input digital signal is divided into a plurality of bands and the tune is determined, fine tune determination can be performed based on the signal components of each band. Further, since the tune determination is performed based on the peak level, it can be configured by a relatively simple circuit.

According to the fourth aspect of the present invention, the encoder performs optimal encoding based on the optimal minimum audibility characteristic and masking characteristic selected according to the tune of the input audio signal.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a signal recording system of an MD recording device according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of a musical composition analysis unit shown in FIG. 1;

FIG. 3 is a diagram illustrating an encoding process by ATRAC when a bit distribution is changed.

FIG. 4 is a diagram showing an encoding process by ATRAC when a bit allocation of a sound group is changed.

FIG. 5 is a diagram illustrating an encoding process by ATRAC when a time width of a sound group is changed.

FIG. 6 shows ATRAC when masking characteristics are changed.
FIG. 7 is a diagram showing an encoding process according to FIG.

FIG. 7 is a diagram illustrating the concept of compression processing in ATRAC.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... MD recording device 8 ... A / D converter 10 ... ATRAC encoder 12 ... Music analysis part 14 ... CPU 16 ... Recording part 18 ... MD disk 20 ... Low pass filter 21,22 ... Band pass filter 23,24,25 ... Level Detection circuit

Claims (4)

[Claims] An encoding unit configured to generate a spectrum signal from an input digital audio signal, and compression-encode the generated spectrum signal based on characteristics of parameters used for compression; A characteristic control unit for determining a tune of a song composed of the signal from the signal, and selecting one of the plurality of characteristics prepared in advance based on the determination result; Is a signal processing device that performs encoding using the characteristic selected by the characteristic control unit. 2. The signal processing apparatus according to claim 1, wherein the characteristic control means continuously determines a tune and selects the characteristic in real time. 3. The characteristic control unit includes: a band dividing unit that divides the input digital audio signal into a plurality of bands; a unit that detects a peak level of a signal in each of the divided bands; The signal processing device according to claim 1, further comprising: a determination unit configured to determine a tune based on the level. 4. The signal processing apparatus according to claim 1, wherein the parameter includes at least one of a minimum audible characteristic and a masking characteristic.